Low-density neutrophils (LDNs) are a subset of neutrophils that remain in the peripheral blood mononuclear cell (PBMC) fraction after density gradient centrifugation. Their density is similar to that of PBMCs, but different from that of normal-density neutrophils (NDNs), which segregate with other polymorphonuclear cells during density centrifugation (1). The origin and possible cellular functions of LDNs are vague and still controversial. Their presence in PBMC layer might be due to increased activation and degranulation since neutrophils are capable to change their buoyancy upon stimulation or inflammatory conditions. On the other hand, their presence might be explained by the change of cell volume which would make LDNs not activated/degranulated neutrophils, but rather primed neutrophils (2). The LDN phenotypical characterizations also continues to be challenging since there is no specific marker that allows to distinguish LDNs from granulocytic myeloid derived suppressive cells (G-MDSC) (3). They are even considered to be the same population by some, and a functional assay to confirm their immunosuppressive properties appears to be a good discriminative option (4).

Low-density neutrophils were first described in systemic lupus erythematosus (SLE) (5), and their count correlates with the severity of vasculitis. Since then, LDNs have been identified in cancer (6, 7), other autoimmune diseases (8–10), infections (8–11), and inflammation (12, 13) where they are speculated to contribute to disease pathology. LDNs have varying functions among inflammatory disorders, indicating the disease-specific LDN diversity. In SLE, LDNs can damage endothelial cells and produce pro-inflammatory cytokines (1), whereas in rheumatoid arthritis (RA), LDNs are not as functionally efficient as NDNs (9). Nevertheless, LDNs are considered highly pro-inflammatory because of their ability to spontaneously produce neutrophil extracellular traps (NETs), reactive oxidative species (ROS), cytokines, and display endothelial toxicity. Their presence is often associated with the severity of several immune-mediated diseases (14). Also, it is possible that LDN functions vary among different diseases. In SLE, LDNs can activate T cells (10), on the other hand, in cancer, LDNs are capable inhibiting T cell responses and promote tumor growth and metastization (15). Because of their immunosuppressive properties, they represent a potential target for immunotherapies focused on disinhibition of the effector cytotoxic T lymphocytes (16).

Juvenile idiopathic arthritis (JIA) is commonly considered to be a multifactorial autoimmune disease driven by a combination of environmental triggers and genetic susceptibility. It is characterized by arthritis that lasts for more than 6 weeks and occurs before 16 years of age. The most common form of JIA is oligoarticular arthritis affecting up to four joints, followed by the polyarticular form (>5 joints), enthesopathy-related arthritis (ERA), systemic JIA (sJIA), and rheumatoid factor (RF)-positive polyarthritis. The JIA subtypes differ not only by clinical symptoms but also by the immunopathogenesis. Most children first manifest the symptoms at around 3 or between 6 and 12 years of age (17–20).

The pathophysiology of JIA is still not completely understood. The key role of neutrophils in the early steps of the pathological immune response has been elucidated (21–25). Neutrophils are the most prevalent cells in the synovial fluid of inflamed joints and exhibit signs of hyperactivation and an impaired capacity to suppress T cell proliferation. In addition, they exhibit defective oxidative bursts and phagocytic activity (21, 22, 26). Notably, using public microarray datasets of PBMCs collected from patients with JIA, a neutrophil activation signature has been identified and the possible involvement of LDNs in JIA pathology has been reported (25).

Therefore, in the present study, we determined the number of LDNs in patients with JIA and focused on their deeper characterization, particularly their phenotypic and functional differences, and their correlation with clinical assessments of patients.

In this study, we identified and characterized LDNs in the PBMCs of patients with JIA using transcriptomic analysis and flow cytometry. LDNs from patients with JIA were phenotypically distinct from NDNs. In addition, LDNs from patients with JIA were increased and showed altered expression of multiple surface molecules compared to LDNs from HDs. Notably, the expression of CD62L, CD66b, and CD11b on the surface of LDNs from patients with JIA was decreased and negatively correlated with the number of inflamed joints and JADAS.

The analysis of publicly available datasets of RNA isolated from PBMCs of patients with JIA indicated the presence of neutrophils in the PBMC fraction (25). In addition, we suspected neutrophil infiltration into the PBMC fraction as the transcription analysis identified neutrophil degranulation as one of the most significantly upregulated processes in PBMCs from patients with JIA. Also, the elevated LDN percentage in JIA has already been reported (25), and although we employed different gating strategies, we arrived at the same conclusion. Neutrophils migrate to the site of inflammation or activate during inflammation, they release biologically active enzymes (28). The levels of these enzymes such as MPO, NE, and MMP8 reportedly increased in the sera of patients with JIA (23). We noted that these enzymes were overproduced in the JIA patients’ PBMC, presumably by LDNs. This is important, because the presence of elevated neutrophil-associated products may contribute to the inflammatory milieu and tissue damage. Moreover, there is evidence of elevated protease levels in adult rheumatoid arthritis (RA) synovial joints (29, 30), therefore a similar pattern may occur in the inflamed joints of patients with JIA.

The origin of LDNs remains unclear. They might represent a completely distinct neutrophil population, as observed in SLE, where they are considered an aberrantly developed lineage of neutrophils resulting from genomic damage (8, 31). In contrast, LDNs may represent neutrophils with the lowest density within a spectrum of buoyant densities, found even in healthy individuals. Inflammatory conditions can lead to a reduction of density of all neutrophils (32, 33). Similarly, neutrophils can shift their density spectrum upon activation (8, 29), and only in cells on the lower end of the density range will occur in the PBMC fraction (34).

The presence of neutrophils in the PBMC fraction has been described in several pathological circumstances; yet, their characterization is inconclusive despite having employed various morphological, functional, and phenotypical markers (35). Both pro- and anti-inflammatory properties and immature and mature morphologies of LDNs have been identified (35). The results of the transcriptomic analysis of LDNs from patients with RA suggested that they resemble immature neutrophils (1, 9). In our hands, cells from JIA patients display distinct immature features. Of note, the immaturity of LDNs is supported by the fact that band immature neutrophils have a lower density (36, 37). The expansion of immature neutrophils has been previously described in patients with JIA (23). Emergency granulopoiesis often occurs during acute and chronic inflammation, leading to the appearance of the immature forms of neutrophils in the circulation. As neutrophil progenitors and early stage neutrophils have high buoyancy, they can occur in the PBMC fraction (38, 39).

The density of healthy mature neutrophils decreases upon in vitro stimulation. Under inflammatory conditions, mature neutrophils can be found in the PBMC fraction, indicating that inflammatory LDNs are mature and activated neutrophils. They often show signs of activation such as increased expression of CD11b and CD66b compared with that in NDNs, as described in patients with advanced adenocarcinoma (40). We observed a decrease in CD62L expression in our JIA cohort, implying an activated state of LDNs in patients with JIA. However, it cannot be excluded that in vivo LDNs are primed and that they subsequently become more susceptible to other stimuli. JIA LDN percentage positively correlated with serum calprotectin levels, a potential marker for diagnosis, prediction of disease relapse, response to treatment and risk of flares in JIA and other autoimmune diseases, such as RA, SLE and inflammatory bowel disease (41–43). Calprotectin is released mostly by phagocyting cells such as neutrophils at the site of inflammation, where it can modulate inflammatory response (44). It is uncertain whether LDNs contribute to the increased calprotectin levels or whether calprotectin activates neutrophils, and they consequently decrease their density and occur in the PBMC fraction. The latter would suggest that systemic inflammation may be the cause of LDN expansion in JIA, however, in our experiments LDN percentage in patients with JIA did not correlate with neither sedimentation rate nor CRP. Degranulation may be another cause of density change. Measurements of granule markers on LDNs, such as CD63, CD11b, and CD66b, supported this concept (40, 45). However, several arguments have been raised against this notion, including the fact that the granular content of LDNs does not demonstrate signs of extensive degranulation (8). In addition, the majority of in vitro activated neutrophils revert to their original density after a couple of hours (46). In this study, LDNs from patients with JIA did not express increased levels of surface granule markers, implying that they had normal granule content.

Neutrophil migration to the sites of infection or injury is a crucial step in innate immunity, and neutrophil retention can lead to chronic inflammation and tissue damage. Activated neutrophils at the sites of inflammation do not necessarily undergo apoptosis and clearance by macrophages and in some circumstances, they may undergo reverse migration and re-enter circulation (reverse transendothelial migration) (47–49). Neutrophils, including their activated and degranulated fractions, arrive at the affected tissues where they perform multiple functions. Neutrophils are the most abundant cells found in synovial fluid of JIA patients and we hypothesized that their LDNs might represent a transmigrated neutrophil subset from the sites of inflammation into the joints (11, 35, 50). Transmigrated neutrophils account for 1–2% of circulating neutrophils in patients with RA, where they produce high amounts of ROS (51). They are hallmarked by a distinctive phenotype of CD54-hi CXCR1-low (51) and decreased expression of CD62L and CD11b (51). The latter two markers were found significantly decreased in the LDNs of patients with JIA in our study, which implies their possible transmigratory origin. Moreover, the expression of CD62L, CD11b, and CD66b negatively correlated with the number of active joints, suggesting that transmigration might be the reason for their decreased expression on the LDN surface in patients with JIA, and that their activated phenotype might be acquired in other tissues. Of note, neutrophils in the tissue may have a phenotype similar to in vitro activated neutrophils (51, 52).

The main limitations in the presented data interpretation are the size and heterogeneity of the studied JIA patient cohort, required for statistical strength to allow robust clinical and cellular phenotype correlations. Four patients were treated with glucocorticoids, which may have affected the neutrophil biology. The data regarding transmigration is indirect, and a phenotypic analysis of transmigrated neutrophils and neutrophils infiltrating the synovial fluid should be performed to confirm the hypothesis of reverse migration of neutrophils. Moreover, functional and morphological studies of LDNs in JIA are lacking. Methodologically, the quantification of degranulation products from Luminex analysis, as well as the analysis of RNA expression performed in PBMC, might be affected by other cells originating from PBMCs.

Several reports have indicated the critical role of neutrophils in JIA immunopathogenesis. Synovial neutrophils exhibit an activated state and impaired functions such as ROS production, phagocytosis, and T cell suppressive features (21, 22). Furthermore, neutrophils in the peripheral blood of patients with JIA demonstrated substantial variations in their subset distribution, phenotype, and a tendency to form aggregates with platelets (PNAs), which affect their activation and pro-inflammatory features (23, 53, 54).

This study contributes to the mounting evidence that LDNs in JIA are altered in their counts, phenotype and functions and may therefore be important players in the disease development or progression. Specifics of their origin, possible contribution to site-specific inflammation, functional characteristics and clinical applications should be addressed by future studies. For instance, the percentage of LDN might be helpful as a biomarker to know subclinical joint inflammation. Larger-sized cohorts should assess whether LDN/LDN features are superior as a marker over other biomarkers, such as calprotectin.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. The data presented in the study are deposited in the GEO repository, accession number GSE235572.

The studies involving humans were approved by the Ethics Committee of University Hospital in Motol. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

ZP designed the study, performed the experiments, statistical and graphical analyses, and wrote the manuscript. IZ performed PBMC culture and nanoString experiments. MB and AK provided healthy donor information, biological material and reviewed the manuscript. RH, HM, and DC provided patient information, biological material and reviewed the manuscript. AS obtained the funding and reviewed the manuscript. All authors contributed to the article and approved the submitted version.